More problems with NLB

[Andy Welcomer]

Hello,

I while back I tried to setup network load balancing on 2 workstations to be
used for testing a production environment. After much headache they worked.
On Friday I got my hands on 2 HP DL380 G3 servers to use for testing. I
have since shutdown the original 2 workstations in favor of setting up the 2
servers for my test. To that end, I used the same names and IPs from the
test environment. However its not working. I get messages in the event log
showing that that the 2 machines have converged together, yet the virtual
doesn't respond. Could this be a result of me trying to use the same names
and IPs again? Any suggestions will be greatly appreciated. Thanks in
advance.

Andy

 

[Chris Darling [MSFT]]

Andy,

There are no dependencies in NLB that would prevent you from reusing server
names or IP addresses. Since the hosts are converged they are able to see
one another, which is a good first step. Try the following to troubleshoot
the problem (I am assuming that you are load balancing HTTP traffic...you'll
need to adjust accordingly if this isn't the case):
1. Open a cmd.exe window on an NLB host and type 'telnet
<cluster-ip-address> 80', to connect to the service through NLB. Do this on
both hosts and make sure that you can connect. If you have connect problems
it could be caused by a) IIS isn't running; b)IIS isn't listening on this
IPort; c) the IP address hasn't been added to TCP/IP on this host. Note
that this test will bypass NLB completely (NLB sits below IP and this test
won't exercise logic below IP).
2. Once that is working try the same thing but from a client that is not on
the NLB hosts. Preferably choose one that is homed to the same LAN. If you
have problems run netmon on client and server while you reproduce the
failure. This should help you figure out the cause. Before you start the
netmon capture, clear the ARP cache on client and server (arp -d from
command line). This way the ARP traffic will be part of the capture.
3. Once that is working repeat the test from a client that is not on the
same LAN as the clients (this introduces a routing layer). If you have
problems repeat the netmon capture process above.

If the above suggestions don't help, check out the NLB troubleshooting guide
(http://www.microsoft.com/technet/prodtechnol/windowsserver2003/technologies/clustering/nlb-trbl.mspx).
If you are still having problems describe the results from the above
troubleshooting steps and post back to this newsgroup. I'll take a look and
offer advice on how to proceed.

Cheers,
Chris

This posting is provided "AS IS" with no warranties, and confers no rights.